Thanks for all the responses. I have since found out that the two
offending machines are PCs running NCSA Telnet v2.3.1 and using
Cabletron E2112 ethernet cards. It appears to be a configuration
problem. The sys admin for the machines is contacting Cabletron (he was
unaware of the problem before I contacted him, but he couldn't get the
NCSA software to address the card properly either).
Most of the responses were that the machines were transmitting the
ethernet broadcast address as their ethernet station address, which
is exactly what was happening, why I don't know. It is possible that the
drivers on the PCs were able to alter the station address.
* MY ORIGINAL POSTING:
I'm seeing the following messages in my system logs:
arp: ether address is broadcast for IP address 80fc7645!
arp: ether address is broadcast for IP address 80fc7646!
These two addresses do exist on our subnet, but I don't know what
machine types they are. I've looked in TFM and cannot find a
discussion of arp errors that covers this (if you know where to look
please enlighten me). Any ideas what's causing these messages?
I'm seeing it on both SPARC1+ and SPARC2 machines with both
SunOS 4.1.1 and 4.1.2.
-------------------------------------------------------------------------
* RESPONSES:
>From barmar@Think.COM Tue Feb 25 00:46:04 1992
Organization: Thinking Machines Corporation, Cambridge MA, USA
I think this means that the machines with those addresses are sending
invalid ARP packets, with ethernet broadcast addresses where the nodes'
ethernet addresses are supposed to be.
Sounds like a bug in the ARP implementation for those two devices.
-- Barry Margolin System Manager, Thinking Machines Corp.barmar@think.com {uunet,harvard}!think!barmar
-------------------------------------------------------------------------
>From katkam@fuwutai.att.com Mon Feb 24 20:07:22 1992
Todd, The messages may be caused if you have duplicate IP addresses for two systems i.e. If you have the same IP address for 2 different systems then there is a possibility any system that is broadcasting and arp ( Address resolution Protocol ) message get a rarp ( reverse address resolution protocol ) message from the client. In this case the client may be both Sparc stations. Since both are responding to that address and give a different ethernet address. This kind of scenario occurs if there is a same IP address for 2 different workstations. On a similar not since your message is concerning different IP address, You may be giving 2 different IP address for the same workstation. I recommend you read the manual (8) on "arp" and look for the -d ( d option ). login to that host ( sparc station ) and check for the /etc/hosts file and look for the host entry address. If you are running NIS look at the master /etc/hosts file entry for that work station.
The problem should be simple. It's only a case of duplicate IP address entry or ethernet address entry. One other place to look at is /etc/ethers file on NIS master and if the client if it is a diskless client.
Anil Katakam AT&T Bell Labs Whippany, NJ, katkam@fuwutai.att.com
-------------------------------------------------------------------------
>From macphed@dvinci.usask.ca Mon Feb 24 18:11:02 1992
Todd;
What is happening is that your computers are seeing packets on the ethernet which have a source address of ff:ff:ff:ff:ff:ff, or in other words, the broadcast address. THIS SHOULD NOT BE HAPPENING!
Why is it happening? I can't say, but it could be hardware problems on these two machines, configuration problems on them, or network problems with corrupted packets.
You might want to use a network "sniffer" to try and grab these packets, then examine them to determine if the packet is corrupted, or if they are being missent.
Ian. Ian MacPhedran, Engineering Computer Centre, University of Saskatchewan. 2B13 Engineering Building, U. of S. Campus, Saskatoon, Sask., CANADA S7N 0W0 macphed@dvinci.USask.CA macphedran@sask.USask.CA
-------------------------------------------------------------------------
>From clyde@sirius.cc.utexas.edu Tue Feb 25 08:42:03 1992
What is probably happening is that those two machines are misconfigured and so are saying that their Ethernet interface hardware addresses are 0xffffffffffff, which is the *hardware* Ethernet broadcast address.
What I would suspect is that those systems have Ethernet interfaces which allow setting of the hardware by host software, and that configuration (host software) is misconfigured in some way.
It should not be fatal to the network but may confuse some hosts on it.
Clyde Hoover |"Personally, I don't see why a guy can't have (Shouter-To-Dead-Parrots) | both a girl and a dog. But if you can only UNIX Sys Admin Tiger Team | afford one of them, get a dog." Computation Center, UT Austin | -Groucho Marx clyde@emx.utexas.edu |
--
------------------------------------------------------------------------- Todd Gamble, Systems Administrator Phone: (314) 362-2011 Washington University School of Medicine FAX: (314) 362-6110 Campus Box 8225 510 South Kingshighway Blvd. St. Louis, Missouri 63110 Email: todd@petadmin.wustl.edu -------------------------------------------------------------------------
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:36 CDT